On 2nd August, I published a blog entitled “The IUCN Red Book lion report is based on supposition and extrapolation”.
In that blog, I mentioned that the IUCN Red List had been eagerly awaited by many but was potentially substantially flawed. I mentioned that the report on the status of Africa’s lions was based on 45 “relatively well monitored populations” and their trends from 1993 to 2014. From those numbers, the report mentioned that lions in those sample populations DECLINED by 66% in western and central Africa (actually, not one single central African population was examined, something the IUCN seems to have overlooked), DECLINED by 59% in eastern Africa and INCREASED by 8% in southern Africa.
I mentioned that we should be careful about such unfounded statements and have a close look at the actual data and the methods used to reach the conclusions as well as all the assumptions made along the way.
The raw data used by the IUCN is published in a paper in the Proceedings of the National Academy of Sciences. Immediately you will see some worrying aspects – of the 45 “relatively well monitored populations”, most are represented by very few data points. In fact, almost half of the populations analysed had four data points or less over the many years of “monitoring”, and 8 only had two data points. Those with the scantiest information included some of the largest populations in the analysis, incorporating those of the Kruger, Masai Mara, Murchison Falls, Chobe, Makgadikgadi and Okavango (accounting for about 5,000 of the total of 7,500 lions in the analysis). Wild populations with relatively good data sets included Serengeti, Ngorongoro, Hwange and Kunene, but populations with the most data were all from small fenced populations in South Africa. This should concern you.
I mentioned in the previous blog that the Red List report based their analysis on trends among those populations from 1993 to 2014 (= 21 years, or three lion generations stipulated by the IUCN as a means of assessing the seriousness of the decline and hence whether lions should be considered endangered). But a simple glance at the data shows that only about six populations had data anywhere near 21 years of monitoring. Not to worry, statistics and models to the rescue. The authors explain that they present “time series data” and used a Bayesian state space model to estimate the growth rate of each population. Bayesian statistics are used to measure the uncertainty of an event, and the probability of any given model is evaluated with how well the predicted data “fits” with the observed data. Also, the probability of the model being correct (of course) depends on the quality of the data, and I contend that for many populations the scarce data available will provide very little confidence that the model is correct, and therefore suggesting the IUCN analysis is flawed.
Let me give you an example. The Okavango population in Botswana was evaluated with only two data points – one from 2003 and the other from 2012 (the last datapoint comes from “unpublished” data). The first showed 1,438 lions and the second 1,107 lions present, and the “model” estimated 1,932 lions in 1993 and 1,050 lions in 2013. Therefore the population had declined by 46% according to the IUCN. I would say that it is almost impossible to verify that assumption. First, with only two data points available, it is not reasonable (though perhaps possible via seductive time series models) to estimate what the population was in 1993, ten years BEFORE the first datapoint was gathered. Second, it is almost impossible to know what happened to the lions in the ten years between the two data points. All sorts of scenarios are possible.
To see what I mean, let’s assume you have a big field behind your house with a population of rabbits. You went out and counted them in 2000 and came up with about 800 rabbits. You then went abroad for many years, and counted them again in 2009 when you came back. This time you found 600 rabbits. You assume that the population is declining, but you could be very wrong. For example, after you left, a combination of diseases, foxes, drought, cold weather etc could have reduced the population to a handful of survivors. Before you counted them for the second time, the population was in fact making a good recovery. So what the 600 rabbits might actually represent is a growing population, not one in decline. And with only those two data points you have no way to draw any conclusions, let alone make assumptions about the size of the rabbit population in 1993 or 2014. Too little data means too many mistakes can be made in attempting to analyse time trends.
For sure, the authors of the PNAS paper were aware of such pitfalls, and they admit that the trends in lion populations examined “are intrinsically imprecise”. Also they said that “the most severely declining populations were the least well monitored” (I would even contest that they knew these populations were declining – see the rabbit example above) and that “accurate estimates of short- to medium-term changes require frequent counts, because time-series data consisting of only two to three surveys can inevitably provide only very weak information on long-term trends”. Strangely, such caveats about the weak information used to evaluate trends are missing from the IUCN report. That should worry you.
Here are some other worrying aspects of the IUCN report:
Inclusion of the fenced, heavily managed, and small populations limited by carrying capacity in (mainly) South Africa’s private and some state-operated reserves. The PNAS paper mentions that these populations “require metapopulation management, euthanasia and contraception and only make limited contributions to … conservation outcomes”. Some of the PNAS paper authors mentioned previously in other publications (e.g. Reintroduction of Top-Order Predators, Eds MW Hayward & MJ Somers, 2009, Blackwell) that these small managed populations do not contribute to wild lion conservation, that they were placed there for tourism and not biodiversity purposes, that these populations lack [genetic] integrity of origins and that individual reserve management regimes do not sufficiently collaborate to consider these lions as any sort of “metapopulation (small populations managed cohesively). The IUCN report largely ignores such concerns, but admits that if these small fenced populations had been excluded from their analysis, the overall outcome of their African lion evaluation would have been very different. So different, in fact, that the decline of African lions would have been close to 50% over three generations, and that the IUCN should therefore have listed all lions, regardless of regions, as endangered.
The IUCN report mentions (twice) that trophy hunting contributes positively to lion conservation if properly managed. The IUCN says “Trophy hunting is carried out in a number of sub-Saharan countries and is considered an important management tool for conserving wild land providing financial resource for Lion conservation for both governments and local communities”. Also – “regulatory measures which reduce the profitability of Lion trophy hunting could have widespread negative impacts for wildlife-based land use, anti-poaching and tolerance of Lion outside protected areas”. Statements like those are not called for in any supposedly unbiased assessment as they are baseless and unsubstantiated.
As I mentioned in my previous blog, the IUCN report was eagerly awaited by organizations like the EU Department of Environment, the US Fish and Wildlife Service, the CITES Secretariat, etc. Such is the value placed on these IUCN listings and assessments. But in this case and perhaps in assessments of other species as well, the IUCN comes up significantly short when details are examined. Simply put, the little hard data available on lion populations does not support the conclusions drawn from time series estimates via a Bayesian state space model. The level of confidence in weak data cannot be improved by models and time series manipulations of hypothetical trends of natural populations that could be subject to unpredictable perturbations due to disease outbreaks, droughts, floods, fires and other adversities.
Nevertheless, I would agree with some of the large-scale trends expressed in the report. Surveys of western African lions have shown that these genetically unique populations are present in very small numbers. And that there is a dearth of information on central African lions, and that eastern African populations are experiencing significant declines for a variety of reasons. But I would strongly disagree with the southern African trends proposed by the IUCN (an increase) as the wild populations on which this supposition is based (Botswana, Kruger, Etosha) are too data deficient to support any conclusions. Because of this, the maintenance of African lions listed as “vulnerable” on the IUCN Red List is questionable – the higher listing of “endangered” should have been adopted at least until better and more reliable information becomes available.
It is also apparent from the available data that we have conveniently and complacently paid very little attention to lions. If we are going to make informed rather than conjectural conservation decisions about this species, far more effort and money is required to get real and accurate information. The PNAS report mentions that lions are difficult to count, but that should motivate us to do better rather than accepting predictions, projections and extrapolations derived from flawed models to guide critical conservation decisions. Simply put, lions deserve better.
Image – coursera.com